Peroxisome proliferator-activated receptor γ-2 (PPARG2) is a ligand-dependent transcriptional factor involved in the pathogenesis of insulin resistance. In the presence of a ligand, PPARG2 associates with co-activators, while it recruits co-repressors (CoRs) in the absence of a ligand. It has been reported that the interaction of liganded PPARG2 with co-activators is regulated by the amino-terminal A/B domain (NTD) via inter-domain communication. However, the role of the NTD is unknown in the case of the interaction between unliganded PPARG2 and CoRs. To elucidate this, total elimination of the influence of ligands is required, but the endogenous ligands of PPARG2 have not been fully defined. PPARG1-P467L, a naturally occurring mutant of PPARG1, was identified in a patient with severe insulin resistance. Reflecting its very low affinity for various ligands, this mutant does not have transcriptional activity in the PPAR response element, but exhibits dominant negative effects (DNEs) on liganded wild-type PPARG2-mediated transactivation. Using the corresponding PPARG2 mutant, PPARG2-P495L, we evaluated the role of the NTD in the interaction between unliganded PPARG2 and CoRs. Interestingly, the DNE of PPARG2-P495L was increased by the truncation of its NTD. NTD deletion also enhanced the DNE of a chimeric receptor, PT, in which the ligand-binding domain of PPARG2 was replaced with that of thyroid hormone receptor β-1. Moreover, NTD deletion facilitated the in vitro binding of nuclear receptor CoR with wild-type PPARG2, mutant P495L, and the PT chimera (PPARG2-THRB). Inter-domain communication in PPARG2 regulates not only ligand-dependent transactivation but also ligand-independent silencing.
Peroxisome proliferator-activated receptor γ-2 (PPARG2) belongs to the nuclear hormone receptor (NHR) superfamily (Desvergne & Wahli 1999, Michalik et al. 2006, Lu & Cheng 2010). Through different promoter usage and splicing, the PPARG gene generates PPARG1, 2, and 3. PPARG2 is predominantly expressed in adipocytes and macrophages, and has profound effects on insulin sensitivity (Desvergne & Wahli 1999). PPARGs heterodimerize with the retinoid X receptors (RXRs) on short DNA sequences, designated the PPAR-responsive element (PPRE). PPRE consists of a pair of directly repeated half sites (typically AGGTCA) spaced with one random nucleotide (DR1; Desvergne & Wahli 1999). Upon ligand binding, the PPARG2-RXR heterodimer recruits co-activators including members of the p160 protein family, CREB-binding protein/p300, and thyroid hormone receptor (THR)-associated protein 220 (Ge et al. 2002), resulting in transactivation (Perissi & Rosenfeld 2005). The function of the ligand-binding domain (LBD) is known to be modulated by the phosphorylation of a serine at codon 112 (S112) in the amino-terminal A/B domain (NTD) by mitogen-activated protein kinase (MAPK; Hu et al. 1996, Adams et al. 1997, Camp & Tafuri 1997, Shao et al. 1998). Based on functional analyses of S112 phosphorylation, Shao et al. (1998) proposed ‘inter-domain communication’, where NTD controls the ligand-binding affinity of the LBD and recruitment of SRC-1, a member of the p160 family.
In the absence of a cognate ligand, THR α-1 (THRA1), β-1 (THRB1), and retinoic acid receptor α (RARA) are known to interact with co-repressors (CoRs) including nuclear receptor CoR (NCoR) and silencing mediator of retinoic acid and THRs (SMRT), resulting in transcriptional silencing. The interaction of CoRs with unliganded PPARG2-LBD has been demonstrated in vitro (Zamir et al. 1997, Krogsdam et al. 2002, Lee et al. 2002, Stanley et al. 2003) and in vivo (Gurnell et al. 2000, Wang et al. 2004, Yu et al. 2005). RNA interference assays against the CoR genes (Guan et al. 2005, Yu et al. 2005) and chromatin immunoprecipitation assays (ChIP) suggest that PPARG2 associates with CoR in vivo and suppresses the basal transcriptional activity of PPARG2-target genes including glycerol kinase (Guan et al. 2005), oxidized low-density lipoprotein receptor 1 (Chui et al. 2005), aP2, and PPARG itself (Picard et al. 2004). However, there is no evidence that the NTD regulates the interaction of its LBD with CoR in the absence of a ligand.
Unlike THRs or RARA, studies using conventional reporter assays have failed to demonstrate the silencing function of unliganded PPARG2 (Zhang et al. 1999, Chatterjee 2001, Semple et al. 2006). In 293EBNA cells without the addition of an exogenous ligand, PPARG did not exhibit silencing function but rather constitutive activation of basal transcription (Barroso et al. 1999, Gurnell et al. 2003, Agostini et al. 2004, Semple et al. 2006). Additionally, overexpression of NCoR did not potentiate the suppression by PPARG2 of acyl-CoA oxidase-derived PPRE (Aox-PPRE) in 293T cells (Zamir et al. 1997). To detect the silencing effect of wild-type PPARG2 in a conventional reporter assay, it is necessary to maintain the majority of PPARG2 molecules in truly an unliganded condition. Endogenous ligands for PPARG2 are reported to be generated by 3T3-L1 pre-adipocytes (Kim et al. 1998, Camp et al. 2001, Madsen et al. 2003, Tzameli et al. 2004) and macrophages (Yano et al. 2007). Similar substances may also be generated by CV1 cells (DiRenzo et al. 1997, Werman et al. 1997), which have been often used in the studies of NHRs. Shi et al. (2002) reported that, in CV1 cells, ‘unliganded’ PPARG2 did not suppress ligand-induced transcription by PPARα (PPARA). These putative ligand(s) for PPARG2 are predicted to dissociate CoR from PPARG2 (Gurnell et al. 2000, Wang et al. 2004, Yu et al. 2005).
Barroso et al. (1999) and Savage et al. (2003) reported two different heterozygous mutations, PPARG1-P467L and V290M, in patients with severe insulin resistance. Both mutants, PPARG1-P467L in particular, have very low affinity for the PPARG ligands and constitutively bind with CoR (Barroso et al. 1999, Agostini et al. 2004). They are believed to be resistant to the endogenous ligand since they exhibited no transactivation in 293EBNA cells in the absence of an exogenous ligand (Barroso et al. 1999, Agostini et al. 2004). These mutant PPARG1s also exhibited potent dominant negative effects (DNEs) on the ligand-induced transactivation by wild-type PPARG1 (Barroso et al. 1999). These properties are reminiscent of the DNE seen in mutant THRB1, identified in patients with resistance to thyroid hormones (RTH; Yoh et al. 1997, Chatterjee 2001), since mutant THRB1s lose their T3-binding activity due to amino acid substitutions in their LBD (Refetoff et al. 1993) and exhibit DNEs by continuous interaction with CoR (Yoh et al. 1997). DNEs against ligand-induced transactivation by wild-type PPARGs have been reported in artificially generated mutant PPARGs that mimic the amino acid substitution of mutant THRB1s in RTH (Zamir et al. 1997, Gurnell et al. 2000, Armoni et al. 2003, Park et al. 2003, Freedman et al. 2005).
Using the analogy of inter-domain communication in ligand-dependent transactivation involving PPARG2, we speculate that the NTD may regulate the interaction of CoR with the unliganded LBD. To confirm this, we took advantage of the fact that PPARG1-P467L has very low affinity for a variety of known ligands (Barroso et al. 1999, Kallenberger et al. 2003, Agostini et al. 2004). We examined the effect of NTD truncation on the DNE of PPARG2-P495L, which possesses an amino acid substitution corresponding to PPARG1-P467L (Barroso et al. 1999, Gurnell et al. 2003, Agostini et al. 2004, Semple et al. 2006). In addition, we generated a chimeric receptor, PT, by fusing the PPARG2-derived NTD and DNA-binding domain (DBD) to the THRB-LBD in order to evaluate its NTD function. Our study demonstrates that the PPARG2-NTD functions to attenuate the association of CoR with unliganded LBD.
Materials and methods
A PT chimera was created by fusing the NTD, DBD, and the A- and T-boxes of mouse PPARG2 (codon 1–232) to the CoR box and LBD of human THRB1 (codon 192–461) using standard molecular techniques. An expression plasmid for mouse PPARG2-P495L was generated from pSG5-mouse PPARG2 using a site-directed mutagenesis kit (Stratagene, La Jolla, CA, USA) in the same codon as used for human patients (Barroso et al. 1999). The construction of mutant THRB1, pCMX-hTHRB1-K443E, and tk-TREp-Luc is described elsewhere (Sasaki et al. 1995). Using the PCR with the PT chimera, PPARG2-P495L, and the THRB1-K443E as templates, dNPT, PPARG2-dNP495L, and THRB1-dNK443E were generated respectively. In dNPT and PPARG2-dNP495L, the N-terminal 127 amino acids were truncated, while, in THRB1-dNK443E, the N-terminal 94 amino acids of THRB1-K443E were deleted. The AHT mutation in the CoR box of the THRB1 hinge domain (codon 263 Ala to Gly, codon 264 His to Gly, and codon 267 Thr to Ala; Horlein et al. 1995) was generated with identical codons in the LBD of PT and dNPT to create PT-AHT and dNPT-AHT respectively. The validity of all plasmid constructs was confirmed by sequencing.
Cell culture and transient transfection
CV1 cells were maintained in DMEM supplemented with 10% FCS. Using the calcium phosphate precipitation method (Sasaki et al. 1995), 6-well dishes were transfected with 300 ng of the luciferase reporter gene containing three copies of rat Aox-derived PPRE fused to thymidine kinase (tk-AoxPPRE-Luc), 1200 ng of β-galactosidase expression vector (a modified version of pCH110), and expression plasmids for receptors in the quantities indicated in the figure legends. Empty vector (pCMX or pSG5) was added to adjust the total amount of DNA. The reporter assay with hADRP-4k-Luc was performed using same protocol as tk-AoxPPRE-Luc. The pre-adipocyte cell line 3T3-L1 was maintained in DMEM with 10% FCS. Using 12-well dishes, the cells were transfected by the lipofection method (Sasaki et al. 1999) with 300 ng of tk-AoxPPRE-LUC reporter gene, plasmids for receptors, and 500 ng of pRL-TK expression vector (Promega) to standardize transfection efficiency. The luciferase activity was normalized using β-galactosidase activity in CV1 cells or by pRL-TK activity in 3T3L1 cells. To evaluate DNE in CV1 and 3T3L1 cells, wild-type receptor plasmid was co-transfected with twofold molar excess of mutant receptor plasmid. All experiments were performed in duplicate and repeated more than three times.
Gel shift assay
Oligonucleotides containing the Aox-derived PPRE (sense; 5′-aggggaccaggacaaaggtcacgttcggg-3′) were labeled with 32P-γ-ATP using T4 polynucleotide kinase (Toyobo, Osaka, Japan). Receptor proteins were translated using the TNT in vitro translation kit (Promega). The translated receptor proteins and 32P-labeled PPRE oligonucleotides (5×104 c.p.m.) were mixed and incubated for 20 min at room temperature. The reaction mixture was loaded onto 5% polyacrylamide gel. The gel was dried and visualized using the BAS-1000 autoradiography system (Fuji, Tokyo, Japan).
GST pulldown assay
Receptor proteins were generated using the TNT in vitro translation kit (Promega) in the presence of 35S-methionine. Bacterially expressed glutathione S-transferase (GST) protein fused to the receptor interacting domain of NCoR (GST–NCoR; Kawai et al. 2004) was immobilized on glutathione-Sepharose 4B beads (Amersham Life Science). The beads were washed once with binding buffer (25 mM HEPES, pH 7.6, 2 mM dithiothreitol, 20 mM NaCl, 20% glycerol, 2 mg/ml BSA, and protease inhibitors (0.3 mM phenyl methylsulfonylfluoride, 1.0 μg/ml leupeptin, and 2.0 μg/ml aprotinin)), before being mixed with binding buffer and left overnight at 4 °C. 35S-methionine-labeled in vitro translated receptor proteins were incubated with the GST fusion protein-bound beads for 3 h at 4 °C. After the incubation, the beads were washed twice with 1 ml of binding buffer +0.1% Triton X-100 and once with 1 ml of 50 mM Tris–HCl (pH 8.0) buffer. The eluted proteins were resolved by SDS-PAGE and visualized by the BAS-1000 autoradiography system (Fuji).
Each experiment was performed in duplicate more than three separate times, and each result is expressed as the mean±s.d. The statistical significance was determined by the ANOVA and Fisher's protected least significant difference (PLSD) test using StatView 4.0 software (Abacus Concepts, Berkeley, CA, USA). P<0.05 was considered significant.
The DNE of PPARG2-P495L is enhanced by the truncation of its NTD
We generated the mutant PPARG2 by deleting its NTD (dNPPARG2; Fig. 1A) and transfecting it into CV1 cells. The expression level of dNPPARG2 was comparable with that of wild-type PPARG2 (Fig. 1B). In the presence of troglitazone, both wild-type PPARG2 and dNPPARG2 stimulated the luciferase reporter gene containing Aox-PPRE fused with the tk promoter (tk-AoxPPRE-Luc; Fig. 1C, lanes 2–5). Although the existence of constitutive activity in the PPARG2-NTD was reported (Werman et al. 1997), truncation of the NTD did not affect transactivation (Fig. 1C, lanes 3 and 5). Without the addition of troglitazone, wild-type PPARG2 and dNPPARG2 showed stimulated basal levels of tk-AoxPPRE-Luc activity (Fig. 1C, lanes 2 and 4), suggesting that they had no silencing effect. To exclude the influence of endogenous ligands contained in the CV1 cells (DiRenzo et al. 1997, Werman et al. 1997), we employed PPARG2-P495L, which has no ligand-binding ability due to its amino acid substitution corresponding to PPARG1-P467L, and compared it with its NTD-truncated mutant, PPARG2-dNP495L (Fig. 1A). The expression levels of PPARG2-P495L and dNP495L were comparable with that of wild-type PPARG2 (Fig. 1B). Both PPARG2-P495L and dNP495L completely lost their ligand-independent (Fig. 1C, lanes 1, 6, and 8) and ligand-dependent transactivation functions (lanes 7 and 9). Co-expression of a twofold excessive amount of PPARG2-P495L repressed the troglitazone-dependent transactivation induced by wild-type PPARG2 (lanes 3 and 10), suggesting the presence of a DNE caused by PPARG2-P495L. Remarkably, the DNE of PPARG2-dNP495L was much more potent than that of PPARG2-P495L (lanes 3, 10, and 11). Likewise, PPARG2-dNP495L exhibited a more rigorous DNE than PPARG2-P495L in 3T3L1 pre-adipocytes (Fig. 1D). Since removal of the NTD potentiates the DNE, the PPARG2-NTD may interfere with the DNE of PPARG2-P495L. We tested the DNE on the activity of Aox-PPRE induced by liganded PPARG1. As shown in Fig. 2A, DNE by PPARG2-dNP495L was stronger than that by P495L (Fig. 2A), although its potency was milder than that on PPARG2-mediated activity (Fig. 1C). A functional PPRE has been reported in the promoters of the human and mouse adipose differentiation-related protein (ADRP) genes (Tachibana et al. 2005). We examined the DNE on the troglitazone-induced activity of the reporter gene, hADRP-4K-Luc, where the human ADRP promoter containing −2981/+1066 was fused with luciferase gene. Ligand-induced activity of the ADRP-derived PPRE by dNPPARG2 (lanes 4 and 5) was similar to that by wild-type PPARG2 (lanes 2 and 3), suggesting that deletion of NTD does not affect the transactivation function of the ADRP gene. While co-expression of twofold excessive amount of P495L did not repress the troglitazone-dependent transactivation induced by wild-type PPARG2, same amount of PPARG2-dNP495L exhibited mild but significant DNE (Fig. 2B), suggesting that NTD truncation potentiates DNE on ligand-induced transactivation via ADRP-derived PPRE.
The DNE of mutant THRB1-K443E is not potentiated by the truncation of its NTD
To compare the effect of the NTD truncation of PPARG2 with that of THRB1, we generated THRB1-dNK443E (Fig. 3A), in which the NTD was deleted from a mutant THRB1-K443E that had been identified in a patient with RTH (Sasaki et al. 1995). The expression levels of THRB1-K443E and THRB1-dNK443E were comparable with that of wild-type THRB1 (Fig. 3B). The luciferase-based reporter gene, which had a typical T3-responsive element (TRE) fused to the tk promoter (tk-TREpal-Luc), was transfected into CV1 cells together with the wild-type and mutant THRB1 expression plasmids. Unliganded THRB1 alone significantly repressed the basal transcriptional activity (Fig. 3C), suggesting transcriptional silencing. As we reported previously (Sasaki et al. 1995), the co-expression of THRB1-K443E reduced the T3-dependent transactivation of tk-TREpal-Luc stimulated by wild-type THRB1, indicating that THRB1-K443E has a DNE on wild-type THRB1. Contrary to the results with PPARG2-P495L (Fig. 1C, lanes 3, 10, and 11), the DNE induced by THRB1-dNK443E was comparable with that of full-length THRB1-K443E, suggesting that NTD truncation does not affect the DNE.
Truncation of the NTD in PPARG2s, but not THRB1-K443E, facilitates the in vitro interaction between its LBD and NCoR
The enhancement of the DNE by the NTD truncation in PPARG2-P495L led us to investigate whether this domain modulates the interaction of its LBD with CoR. We performed an in vitro binding assay using bacterially expressed NCoR fused to GST (GST–NCoR) and 35S-radiolabeled receptors. Deletion of the NTD from PPARG2-P495L drastically increased its binding affinity to NCoR (Fig. 4A, lanes 8 and 9). We tested another mutant PPARG2, PPARG2-V318M. This mutant PPARG2 possesses an amino acid substitution corresponding to PPARG1-V290M (Barroso et al. 1999), LBD of which has the structure different from that of PPARG1-P467L (Kallenberger et al. 2003). Although PPARG2-V318M may have some affinity for PPARG ligands (Agostini et al. 2004), in vitro experiments allowed us to evaluate the NCoR interaction in the ligand-free condition. As shown in Fig. 4B, the NTD-deleted mutant, PPARG2-dNV318M, also exhibited increased affinity for NCoR (lanes 5 and 6). The in vitro translated products of THRB1 and THRB1-K443E produced two major bands due to the two initiation methionines in their NTDs (Fig. 4C, lanes 1 and 2), while THRB1-dNK443E appeared as a single band (lane 3). Although we utilized an excessive amount of GST–NCoR and adjusted the amounts of radiolabeled receptors in the input lanes (lanes 1–3), the signals from discrete bands were not affected by truncation of the NTD (lanes 8 and 9). Because there is the possibility that structure of CoR-binding surface may be altered in mutant PPARG2s, we tested the wild-type PPARG2 and dNPPARG2. As shown in Fig. 4D (lanes 5 and 6), the interaction of wild-type PPARG2 with NCoR was again enhanced by the truncation of its NTD. Together, it is considered that the NTD in PPARG2s, but not THRB1-K443E, is able to prevent the interaction of NCoR with its LBD.
The construction of the chimeric receptor, PT, in which PPARG2-derived NTD and DBDs were fused to a THRB1-derived LBD
To exclude the influence of endogenous ligand in CV1 and 3T3L1 cells, we further attempted to generate a chimeric receptor, in which the PPARG2-LBD was replaced with a THRB1-derived LBD. The PPAR–RXR heterodimer binds to a DR1-type PPRE with PPAR on the 5′ end and RXR on the 3′ end (Desvergne & Wahli 1999). The inverted polarity is dictated by the A- and T-boxes in the PPARG2-DBD. Based on these considerations, we constructed a chimeric receptor designated PT, which possesses A- and T-boxes derived from PPARG2 (Fig. 5A). A gel shift assay revealed that the PT chimera, as well as PPARG2, formed heterodimers with RXR on PPRE (Fig. 5B). As expected, deletion of the A- or T-box abolished heterodimer formation with RXR (data not shown). As shown in Fig. 4C, the PT chimera activated tk-AoxPPRE-Luc in the presence of 100 nM T3 to a level similar to that induced by wild-type PPARG2 in the presence of 5 μM troglitazone. In contrast with wild-type PPARG2, the PT chimera significantly repressed basal transcriptional activity (Fig. 5C, inset), suggesting that PT possesses a silencing function in the absence of T3. The PT chimera did not stimulate tk-TREpal-Luc, presumably due to the specificity of the PPARG2-derived DBD (Fig. 5D).
The truncation of NTD potentiates the DNE of the PT chimera receptor via interaction with NCoR
We tested whether the PT chimera in the absence of T3 also interferes with the troglitazone-induced transactivation driven by wild-type PPARG2. In CV1 cells, transactivation of tk-AoxPPRE-Luc driven by liganded wild-type PPARG2 was inhibited by the co-expression of the PT chimera in the absence of T3 (Fig. 6B, lanes 3 and 4). The CoR box in the THRB1 hinge region (Horlein et al. 1995) is necessary for interaction with CoRs (Marimuthu et al. 2002). To investigate whether the DNE of the PT chimera is mediated through its association with CoRs, we introduced the AHT mutation (Horlein et al. 1995) into the CoR box of the PT chimera (Fig. 6A). As illustrated in Fig. 6B (lanes 4 and 6), the DNE of PT-AHT was abrogated, indicating that the repression by the PT chimera was mediated through its interaction with CoRs. Subsequently, we constructed dNPT, in which the NTD was deleted from the PT chimera (Fig. 6A). While the T3-dependent transactivation by dNPT was similar to that of the PT chimera (Fig. 6C), the DNE of dNPT was more potent than that of the full-length PT chimera (Fig. 6B, lanes 4 and 5). The introduction of the AHT mutation into the CoR box of dNPT again reduced its DNE (Fig. 6B, lanes 5 and 7). The expression levels of dNPT, PT-AHT, and dNPT-AHT were comparable with that of the PT chimera (Fig. 6D). Similar results were obtained in 3T3L1 pre-adipocytes (Fig. 6E). As shown in Fig. 6F, the in vitro interaction with NCoR was enhanced by the truncation of the NTD in the PT chimera. These results indicate that PPARG2-NTD inhibits the binding of CoRs with the LBD derived from PPARG2 as well as that derived from THRB1, resulting in transcriptional upregulation.
It was reported that the association of CoRs requires the CoR box, which is typically found in the hinge region of THRs and RARA (Horlein et al. 1995). However, later analyses revealed that the amino acid residues directly interacting with CoRs are distributed widely in THR-LBD, although the CoR box is necessary for maintaining the global structure of the THR-LBD for CoR interaction (Marimuthu et al. 2002). This implies that the CoR box is not necessarily the determinant of CoR interaction. X-ray crystal analysis showed that CoRs interact with the PPARA-LBD (Xu et al. 2002, Agostini et al. 2004), the interaction surface of which is very similar to that of the THR-LBD (Kallenberger et al. 2003). PPARA has high homology (∼71%) in the amino acid sequence of its LBD with the PPARG-LBD (Xu et al. 2002, Agostini et al. 2004). Direct binding of CoRs with wild-type PPARG2 was demonstrated by GST pulldown assay (Zamir et al. 1997, Krogsdam et al. 2002, Stanley et al. 2003), co-immunoprecipitation (Gurnell et al. 2000, Yu et al. 2005), gel shift assay (Lee et al. 2002), and an in vitro pulldown assay with RXR-PPARG2 heterodimers on PPREs (Krogsdam et al. 2002). Using a ChIP assay, Guan et al. (2005) showed that, in adipocytes without the addition of an exogenous ligand, PPARG2 molecules associated with CoRs are present in PPREs in the promoter region of the glycerol kinase gene (Guan et al. 2005). Similar results were reported in other PPARG-target genes including oxidized low-density lipoprotein receptor 1 (Chui et al. 2005), aP2, and PPARG (Picard et al. 2004). These findings suggest that unliganded PPARG2 recruits CoRs, resulting in transcriptional silencing. It should be noted, however, that the NCoR recruitment may depend on the sequence of PPREs (Guan et al. 2005). This could explain why twofold excessive amount of PPARG2-P495L failed to exhibit DNE on the ADRP promoter activity (Fig. 2B).
PPARG-LBD possesses a very large ligand-binding pocket (1300 A3; Desvergne & Wahli 1999) and can accommodate a variety of ligands (Michalik et al. 2006, Semple et al. 2006). It is currently difficult to completely eliminate the endogenous ligands produced by cells (DiRenzo et al. 1997, Werman et al. 1997, Kim et al. 1998, Camp et al. 2001, Madsen et al. 2003, Tzameli et al. 2004, Yano et al. 2007). The existence of endogenous ligands may mask the silencing by unliganded PPARG2. This could explain why conventional reporter assays have failed to detect the silencing effect by PPARG2 despite the absence of exogenous ligands (Zamir et al. 1997, Barroso et al. 1999, Zhang et al. 1999, Chatterjee 2001, Gurnell et al. 2003, Agostini et al. 2004, Semple et al. 2006). For the creation of ligand-free conditions in vivo, we took advantage of the fact that PPARG1-P467L is resistant to various ligands (Barroso et al. 1999, Agostini et al. 2004) and generated a corresponding mutant, PPARG2-P495L. Our analyses with a conventional luciferase reporter assay demonstrated that NTD truncation potentiated the DNE of PPARG2-P495L on the ligand-induced transactivation induced by wild-type PPARG2 (Fig. 1C and D). Potentiation of DNE by NTD truncation was also observed in the PPARG1-mediated activation of Aox-PPRE (Fig. 2A) and in the transactivation of the human ADRP promoter stimulated by PPARG2 (Fig. 2B). Moreover, in the cell-free conditions with GST pulldown assay, truncation of NTD enhances NCoR binding with not only PPARG2-P495L (Fig. 4A) and V318M (Fig. 4B) but also wild-type PPARG2 (Fig. 4D). These results indicate that the NTD controls the association of CoR with unliganded PPARG2-LBD via inter-domain communication, which was originally proposed in the context of ligand-dependent transactivation by PPARG2 (Shao et al. 1998). It is necessary in future to examine whether NTD modulates the effect of ligand to dissociate NCoR from LBD.
We created a chimera, PT, by replacing PPARG-LBD with THRB-LBD. This chimeric receptor has many advantages for analyzing the function of unliganded PPARG2. First, the ligand-binding pocket of THRB-LBD (600 A3) is much smaller than that of PPARGs (Desvergne & Wahli 1999) and is specific for T3. The concentration of T3 in culture medium is easy to measure, and the method for its removal is established (Samuels et al. 1979). Second, the interaction of THRB-LBD with CoR has been characterized in detail (Horlein et al. 1995, Yang et al. 1999, Marimuthu et al. 2002). Accumulated information from the natural THRB1 mutants in RTH is also available (Refetoff et al. 1993). The functional relevance in the context of DNEs is confirmed in mutant PPARGs harboring artificial amino acid substitutions corresponding to those of mutant THRB1s in vitro (Zamir et al. 1997, Gurnell et al. 2000, Chatterjee 2001, Nugent et al. 2001, Park et al. 2003) and in vivo (Freedman et al. 2005). Finally, THR-LBD is flexible with regard to the orientation of its DBD in a variety of TREs (Kurokawa et al. 1993). This feature is critical for mimicking the inverted polarity of the PPARG2-RXR heterodimer in the PPRE (Desvergne & Wahli 1999). Using the PT chimera, we observed that NTD truncation enhanced the DNE on the ligand-dependent transactivation induced by wild-type PPARG2. This suggests that the PPARG2-NTD functions to regulate the association of CoR with unliganded LBD. The data from the AHT mutation in CoR box (Horlein et al. 1995) of the PT chimera indicate that DNEs are mediated through CoR recruitment but not the RXR squelching proposed previously (Juge-Aubry et al. 1995, Hunter et al. 1996, Winrow et al. 1996, Lu & Cheng 2010).
PPARG2-NTD may have multiple functions including direct interactions with p300 (Gelman et al. 1999) and Tip60 (van Beekum et al. 2008) and modification by small ubiquitin-related modifier-1 (Ohshima et al. 2004, Yamashita et al. 2004). Consistently, a microarray analysis using a chimera between PPARG2 and PPARδ (PPARD) revealed that the PPARG2-NTD is the major determinant for the PPARG2-specific gene expression profile (Hummasti & Tontonoz 2006). However, Bugge et al. (2009) recently reported that, of the 257 genes induced by rosiglitazone, only 25 genes exhibited the reduced transcriptional activity in the cell expressing truncated PPARG2 lacking NTD. We speculate that the NTD-dependent function of unliganded PPARG2 to regulate the CoR association may be another determinant for the PPARG2-specific gene expression profile. An attempt to confirm the in vivo function of NTD in unliganded PPARG2 is currently underway in our laboratory. Our study indicates that the inter-domain communication between NTDs and LBDs, originally proposed in the context of ligand-dependent transactivation (Shao et al. 1998), also plays a role in CoR association and the silencing function mediated by unliganded PPARG2. Similar modification of LBD–CoR interactions by the NTD has been suggested in other NHRs (Kumar & Thompson 2003, Pippal & Fuller 2008). For example, insulin-induced phosphorylation in the PPARA-NTD enhances the dissociation of CoRs from its LBD (Juge-Aubry et al. 1999). Direct interaction between NTDs and LBDs was reported in estrogen receptor (Metivier et al. 2000, Metivier et al. 2001) and androgen receptor (AR; Cheng et al. 2002, He & Wilson 2002, Liao et al. 2003). In AR, the interaction between NTD and LBD, referred to as ‘N/C interaction’ (He & Wilson 2002), was interfered by the overexpression of NCoR or SMRT (Liao et al. 2003). It is also necessary to confirm the inter-domain communication of unliganded PPARG2 using the chromatinized PPREs.
Hollenberg et al. (1996) reported that pituitary-specific THRB2 does not have a silencing function in the absence of T3. One may assume common features between the THRB2-NTD and PPARG2-NTD in the mechanism of loss of silencing function. However, in contrast to PPARG2 (Fig. 4D), the truncation of the NTD from THRB2 did not affect its physical interaction with CoRs in vitro (Yang et al. 1999). Thus, the mechanism underlying the CoR interaction of THRB2 is different. According to Yang et al. (1999), CoRs interact with not only THRB2-LBD but also its NTD, and CoR-NTD contact neutralizes the repressive function of CoRs. On the other hand, the same authors reported that, in wild-type THRB1, NTD truncation did not affect its silencing function or in vitro binding with CoR (Yang et al. 1999). This is in agreement with our finding that deletion of the NTD from THRB1-K443E did not affect its interaction with NCoR (Fig. 4C) or its DNE on the transactivation induced by T3-bound THRB1 (Fig. 3C).
Given that the transcriptional silencing by unliganded PPARG2 is controlled by its NTD, genetic modification of this domain may alter its insulin sensitivity in vivo. In the PPARG2-NTD, the serine residue at codon 112 is phosphorylated by MAPK (Hu et al. 1996, Adams et al. 1997, Camp & Tafuri 1997). Rangwala et al. (2003) generated knock-in mice, in which the serine at codon 112 was substituted for phosphorylation-resistant alanine (S112A), and noted an increase in insulin sensitivity. Of note, thiazolidinedione-sensitive genes such as phosphoenolpyruvate carboxykinase (PEPCK) were not significantly activated in S112A mice. Rangwala et al. (2003) suggested that the property of S112A-PPARG2 in the reduction of insulin resistance may be qualitatively different from that of liganded PPARG2. In agreement with this, Armoni et al. (2003) demonstrated in vitro that S112A substitution diminishes the silencing activity caused by unliganded PPARG2. Polymorphism in the NTD of the PPARG2 gene from CCA (proline) at codon 12 to GCA (alanine) (P12A) is also associated with higher insulin sensitivity (Deeb et al. 1998). However, it was shown that transactivation by PPARG2-P12A in the presence of a cognate ligand is similar or reduced compared to that of the wild-type receptor (Deeb et al. 1998, Masugi et al. 2000). Further studies are required to clarify whether these genetic modifications in the NTD may affect the insulin sensitivity through the modulation of the interaction between CoRs and unliganded PPARG2.
Declaration of interest
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.
This work was supported, in part, by a Health Sciences Research Grant to H Nakamura and a Grant-in Aid for Scientific Research to H Morita, S Sasaki, and H Nakamura from the Ministry of Education, Culture, Sports, Science, and Technology in Japan (grant number 13671181).
We are grateful to the following researchers for providing the plasmids: Ronald M Evans (Howard Hughes Medical Institute, The Salk Institute for Biological Studies, USA) for pCMX-hTHRB1 and tk-TREp-Luc; Akira Kakizuka (Kyoto University, Japan) for pCH111; Toshiya Tanaka (University of Tokyo, Japan) for hADRP-4k-Luc; and Steven A Kliewer (Glaxo Wellcome Research and Development) for pSG5-PPARG2 and tk-AoxPPRE-Luc.
AdamsMReginatoMJShaoDLazarMAChatterjeeVK1997Transcriptional activation by peroxisome proliferator-activated receptor gamma is inhibited by phosphorylation at a consensus mitogen-activated protein kinase site. Journal of Biological Chemistry2725128–5132.doi:10.1074/jbc.272.43.27295.
AgostiniMGurnellMSavageDBWoodEMSmithAGRajanayagamOGarnesKTLevinsonSHXuHESchwabeJW2004Tyrosine agonists reverse the molecular defects associated with dominant-negative mutations in human peroxisome proliferator-activated receptor gamma. Endocrinology1451527–1538.doi:10.1210/en.2003-1271.
ArmoniMKritzNHarelCBar-YosephFChenHQuonMJKarnieliE2003Peroxisome proliferator-activated receptor-gamma represses GLUT4 promoter activity in primary adipocytes, and rosiglitazone alleviates this effect. Journal of Biological Chemistry27830614–30623.doi:10.1074/jbc.M304654200.
DiRenzoJSoderstromMKurokawaROgliastroMHRicoteMIngreySHorleinARosenfeldMGGlassCK1997Peroxisome proliferator-activated receptors and retinoic acid receptors differentially control the interactions of retinoid X receptor heterodimers with ligands, coactivators, and corepressors. Molecular and Cellular Biology172166–2176.
GurnellMWentworthJMAgostiniMAdamsMCollingwoodTNProvenzanoCBrownePORajanayagamOBurrisTPSchwabeJW2000A dominant-negative peroxisome proliferator-activated receptor gamma (PPARgamma) mutant is a constitutive repressor and inhibits PPARgamma-mediated adipogenesis. Journal of Biological Chemistry2755754–5759.doi:10.1074/jbc.275.8.5754.
HunterJKassamAWinrowCJRachubinskiRACaponeJP1996Crosstalk between the thyroid hormone and peroxisome proliferator-activated receptors in regulating peroxisome proliferator-responsive genes. Molecular and Cellular Endocrinology116213–221.doi:10.1016/0303-7207(95)03717-9.
Juge-AubryCEGorla-BajszczakAPerninALembergerTWahliWBurgerAGMeierCA1995Peroxisome proliferator-activated receptor mediates cross-talk with thyroid hormone receptor by competition for retinoid X receptor. Possible role of a leucine zipper-like heptad repeat. Journal of Biological Chemistry27018117–18122.doi:10.1074/jbc.270.30.18117.
Juge-AubryCEHammarESiegrist-KaiserCPerninATakeshitaAChinWWBurgerAGMeierCA1999Regulation of the transcriptional activity of the peroxisome proliferator-activated receptor alpha by phosphorylation of a ligand-independent trans-activating domain. Journal of Biological Chemistry27410505–10510.doi:10.1074/jbc.274.15.10505.
KrogsdamAMNielsenCANeveSHolstDHelledieTThomsenBBendixenCMandrupSKristiansenK2002Nuclear receptor corepressor-dependent repression of peroxisome-proliferator-activated receptor delta-mediated transactivation. Biochemical Journal363157–165.doi:10.1042/0264-6021:3630157.
KurokawaRYuVCNaarAKyakumotoSHanZSilvermanSRosenfeldMGGlassCK1993Differential orientations of the DNA-binding domain and carboxy-terminal dimerization interface regulate binding site selection by nuclear receptor heterodimers. Genes and Development71423–1435.doi:10.1101/gad.7.7b.1423.
LeeGElwoodFMcNallyJWeiszmannJLindstromMAmaralKNakamuraMMiaoSCaoPLearnedRM2002T0070907, a selective ligand for peroxisome proliferator-activated receptor gamma, functions as an antagonist of biochemical and cellular activities. Journal of Biological Chemistry27719649–19657.doi:10.1074/jbc.M200743200.
MadsenLPetersenRKSorensenMBJorgensenCHallenborgPPridalLFlecknerJAmriEKriegPFurstenbergerG2003Adipocyte differentiation of 3T3-L1 preadipocytes is dependent on lipoxygenase activity during the initial stages of the differentiation process. Biochemical Journal375539–549.doi:10.1042/BJ20030503.
MasugiJTamoriYMoriHKoikeTKasugaM2000Inhibitory effect of a proline-to-alanine substitution at codon 12 of peroxisome proliferator-activated receptor-gamma 2 on thiazolidinedione-induced adipogenesis. Biochemical and Biophysical Research Communications268178–182.doi:10.1006/bbrc.2000.2096.
MetivierRPenotGFlouriotGPakdelF2001Synergism between ERalpha transactivation function 1 (AF-1) and AF-2 mediated by steroid receptor coactivator protein-1: requirement for the AF-1 alpha-helical core and for a direct interaction between the N- and C-terminal domains. Molecular Endocrinology151953–1970.doi:10.1210/me.15.11.1953.
NugentCPrinsJBWhiteheadJPSavageDWentworthJMChatterjeeVKO'RahillyS2001Potentiation of glucose uptake in 3T3-L1 adipocytes by PPAR gamma agonists is maintained in cells expressing a PPAR gamma dominant-negative mutant: evidence for selectivity in the downstream responses to PPAR gamma activation. Molecular Endocrinology151729–1738.doi:10.1210/me.15.10.1729.
SavageDBTanGDAceriniCLJebbSAAgostiniMGurnellMWilliamsRLUmplebyAMThomasELBellJD2003Human metabolic syndrome resulting from dominant-negative mutations in the nuclear receptor peroxisome proliferator-activated receptor-gamma. Diabetes52910–917.doi:10.2337/diabetes.52.4.910.
TachibanaKKobayashiYTanakaTTagamiMSugiyamaAKatayamaTUedaCYamasakiDIshimotoKSumitomoM2005Gene expression profiling of potential peroxisome proliferator-activated receptor (PPAR) target genes in human hepatoblastoma cell lines inducibly expressing different PPAR isoforms. Nuclear Receptor33doi:10.1186/1478-1336-3-3.
TzameliIFangHOlleroMShiHHammJKKievitPHollenbergANFlierJS2004Regulated production of a peroxisome proliferator-activated receptor-gamma ligand during an early phase of adipocyte differentiation in 3T3-L1 adipocytes. Journal of Biological Chemistry27936093–36102.doi:10.1074/jbc.M405346200.
WermanAHollenbergASolanesGBjorbaekCVidal-PuigAJFlierJS1997Ligand-independent activation domain in the N terminus of peroxisome proliferator-activated receptor gamma (PPARgamma). Differential activity of PPARgamma1 and -2 isoforms and influence of insulin. Journal of Biological Chemistry27220230–20235.doi:10.1074/jbc.272.32.20230.
WinrowCJKassamAMiyataKSMarcusSLHunterJCaponeJPRachubinskiRA1996Interplay of the peroxisome proliferator-activated receptor and the thyroid hormone receptor-signaling pathways in regulating peroxisome proliferator-responsive genes. Annals of the New York Academy of Sciences804214–230.doi:10.1111/j.1749-6632.1996.tb18618.x.
YamashitaDYamaguchiTShimizuMNakataNHiroseFOsumiT2004The transactivating function of peroxisome proliferator-activated receptor gamma is negatively regulated by SUMO conjugation in the amino-terminal domain. Genes to Cells91017–1029.doi:10.1111/j.1365-2443.2004.00786.x.
YangZHongSHPrivalskyML1999Transcriptional anti-repression. Thyroid hormone receptor beta-2 recruits SMRT corepressor but interferes with subsequent assembly of a functional corepressor complex. Journal of Biological Chemistry27437131–37138.doi:10.1074/jbc.274.52.37131.
YanoMMatsumuraTSenokuchiTIshiiNMurataYTaketaKMotoshimaHTaguchiTSonodaKKukidomeD2007Statins activate peroxisome proliferator-activated receptor gamma through extracellular signal-regulated kinase 1/2 and p38 mitogen-activated protein kinase-dependent cyclooxygenase-2 expression in macrophages. Circulation Research1001442–1451.doi:10.1161/01.RES.0000268411.49545.9c.
YuCMarkanKTempleKADeplewskiDBradyMJCohenRN2005The nuclear receptor corepressors NCoR and SMRT decrease peroxisome proliferator-activated receptor gamma transcriptional activity and repress 3T3-L1 adipogenesis. Journal of Biological Chemistry28013600–13605.doi:10.1074/jbc.M409468200.